System and method for modeling complex systems with distributed actor-based simulation
Fully integrated collection of relevant data, analysis of that data and generation of both analysis-driven decisions and analysis driven simulations of alternate candidate actions. This organizational operating system may be used predict the outcome of enacting candidate decisions based upon past and current data retrieved from both within the corporation and from a plurality of external sources pre-programmed into the system. Simulations using this data and predefined parameters to create models of actors are then run. Risk to value estimates of candidate decisions are also calculated.
1 . A distributed system for modeling of complex systems with large and complex datasets using a distributed simulation engine comprising:
a plurality of distributed computing devices connected over a network, with each distributed computing device comprising at least a processor, a memory, and a network interface;
wherein a plurality of programming instructions stored in one or more of the memories and operating on one or more of the processors of the plurality of distributed computing devices causes the plurality of distributed computing devices to:
automatically retrieve or receive a plurality of data from a plurality of sources over the network, wherein the sources include sensors and network devices that provide streaming data in near real-time;
create a world model for a simulation using parameters and the data from the plurality of sources, wherein the world model is configured to represent an aspect of the cyber-physical world, and wherein the world model comprises:
a plurality of constraints establishing boundary conditions matching those of the cyber-physical world under expected conditions of the simulation;
a plurality of actor models, wherein each actor model is configured to act independently of all other actor models within the boundary conditions, and wherein the actor models are configured to operate in parallel within the world model during operation of the world model; and
an environment monitoring layer configured to:
monitor each actor model within the world model during operation of the world model, wherein monitoring the actor models includes monitoring actions of the actor models, interactions between the actor models, and a state of the actor models at multiple points in simulated time;
store the monitored actions, interactions, and state of one or more of the actor models using graph-based information storage, and
pass control directives from the simulation to one or more of the actor models during operation of the world model;
run the simulation until a simulation result is achieved, wherein the simulation is structured to run multiple alternative decision pathways in parallel to generate multiple predictive outcomes;
while the simulation is running, store decision points and their corresponding parallel decision pathways using the graph-based information storage as they are generated, wherein each stored decision point includes a sequence, a context, and a state for the actions of each actor model and interactions among the actor models; and
generate one or more value-at-risk estimations based on the multiple alternative decision pathways by performing a plurality of statistical data analyses on the stored decision points and their corresponding decision pathways in the graph-based information storage based on the simulation parameters and the simulation result.
2 . The distributed system of claim 1 , wherein the statistical data analyses comprise information theory based statistical analysis.
3 . The distributed system of claim 1 , wherein the statistical data analyses comprise Monte Carlo heuristic model value at risk principles.
4 . The distributed system of claim 1 , wherein the distributed system is configured to allow jobs that run in a single iteration with a single set of parameters and jobs that include multiple iterations and sets of predetermined sets of parameters with termination criteria to stop execution when a desired simulation result is obtained.
5 . The distributed system of claim 4 , wherein some jobs are run offline in a batch mode and other jobs are run online in an interactive mode.
6 . The distributed system of claim 1 , wherein the simulation includes models for hazards, vulnerabilities, contractual obligations and financial capital loss.
7 . The distributed system of claim 1 , wherein the simulation result is used as feedback to a subsequently run simulation.
8 . A computer-implemented method for modeling of systems with large and complex datasets using a distributed simulation engine on a plurality of distributed computing devices comprising the steps of:
automatically retrieving or receiving a plurality of data from a plurality of sources over a network, wherein the sources include sensors and network devices that provide streaming data in near real-time;
creating a world model for a simulation using parameters and the data from the plurality of sources, wherein the world model is configured to represent an aspect of the cyber-physical world, and wherein the world model comprises:
a plurality of constraints establishing boundary conditions matching those of the cyber-physical world under expected conditions of the simulation;
a plurality of actor models, wherein each actor model is configured to act independently of all other actor models within the boundary conditions, and wherein the actor models are configured to operate in parallel within the world model during operation of the world model; and
an environment monitoring layer configured to;
monitor each actor model within the world model during operation of the world model, wherein monitoring the actor models includes monitoring actions of the actor models, interactions between the actor models, and a state of the actor models at multiple points in simulated time;
store the monitored actions, interactions, and state of one or more of the actor models using graph-based information storage actions and interactions, and
pass control directives from the simulation to one or more of the actor models during operation of the world model;
running the simulation until a simulation result is achieved, wherein the simulation is structured to run multiple alternative decision pathways in parallel to generate multiple predictive outcomes;
while the simulation is running, storing decision points and their corresponding parallel decision pathways using the graph-based information storage as they are generated, wherein each stored decision point includes a sequence, a context, and a state for the actions of each actor model and interactions among the actor models; and
generating one or more value-at-risk estimations based on the multiple alternative decision pathways by performing a plurality of statistical data analyses on the stored decision points and their corresponding decision pathways in the graph-based information storage based on the simulation parameters and the simulation result.
9 . The method of claim 8 , wherein the statistical data analyses comprise information theory based statistical analysis.
10 . The method of claim 8 , wherein the statistical data analyses comprise Monte Carlo heuristic model value at risk principles.
11 . The method of claim 8 , further comprising the step of allowing jobs that run in a single iteration with a single set of parameters and jobs that include multiple iterations and sets of predetermined sets of parameters with termination criteria to stop execution when a desired simulation result is obtained.
12 . The method of claim 11 , wherein some jobs are run offline in a batch mode and other jobs are run online in an interactive mode.
13 . The method of claim 8 , wherein the simulation includes models for hazards, vulnerabilities, contractual obligations and financial capital loss.
14 . The method of claim 8 , further comprising feeding back the analysis simulation result as input to a subsequently run simulation.
15 . A computer-readable, non-transitory medium comprising a plurality of programming instructions that, when operating on the plurality of distributed computing devices each comprising at least a processor, a memory, and a network interface, cause the plurality of distributed computing devices to carry out the method of claim 8 .